| 06:48 | Dashiva | wonders how othermaciej handled the Garret-Kris spiral of misconceptions |
| 06:54 | <jruderman> | Hixie: i have some questions about https://bugzilla.mozilla.org/show_bug.cgi?id=450981 |
| 06:55 | <jruderman> | Hixie: when the spec says 'meta elements must be inside a head element', does that mean user agents are supposed to ignore them when they appear elsewhere? |
| 06:55 | <jruderman> | Hixie: can you cook up some stats on how many sites would break if we made that change? |
| 06:56 | <Hixie> | when the spec says "meta elements must be inside a head element" it only means that head elements, to be conforming, have to have a head element ancestor |
| 06:56 | <Hixie> | it doesn't mean anything to do with how browsers work |
| 06:59 | <jruderman> | thanks |
| 07:00 | <Hixie> | commented |
| 07:00 | <Hixie> | and invalidated |
| 07:01 | <jruderman> | Hixie: it was originally filed as an anti-XSS measure, not an HTML5 conformance thing |
| 07:01 | <jruderman> | Hixie: but i guess HTML5 actually requires not fixing it |
| 07:02 | <Hixie> | oh, the subject was changed? man i hate it when people do that :-) |
| 07:02 | <Hixie> | but yeah, technically html5 requires it to work anywhere |
| 07:02 | <Hixie> | maybe that should be changed |
| 07:02 | <Hixie> | though i know that at least for the character encoding, we have to support it anywhere |
| 07:03 | <jruderman> | if evil people can screw with character encoding they can probably sneak an XSS in through bytes that don't look like '<' but turn into '<' |
| 07:04 | <jruderman> | so you've seen a number of documents with meta charset in <body>? |
| 07:05 | <Hixie> | meta charset in <body> happens all the time |
| 07:05 | <jruderman> | fun |
| 07:05 | <Hixie> | thing is the <body> starts as soon as you see any character |
| 07:05 | <Hixie> | so e.g. in <html>a<head><meta charset=""></head><body>...</body></html> the meta is in the body |
| 07:06 | <jruderman> | interesting. i've seen pages like that. |
| 07:06 | <jruderman> | i've also gotten frustrated by how whitespace seems to move around near the head ;) |
| 07:07 | <Hixie> | yeah |
| 07:09 | <jruderman> | do you want to comment in the bug again, or should i? it should probably be reopened with the summary reverted, with a suggestion of WONTFIX |
| 07:09 | <Hixie> | i can |
| 07:10 | <jruderman> | thanks |
| 07:10 | <Hixie> | done |
| 07:13 | jruderman | changes the summary again |
| 07:53 | <hdworak> | hi. does html5lib handle CDATA? |
| 07:54 | <hdworak> | I mean CDATA sections, like <![CDATA[<sender>John Smith</sender>]]> |
| 08:37 | <othermaciej> | Dashiva: at some point, I stopped reading |
| 08:55 | <Hixie> | hdworak: HTML5 doesn't have <![CDATA[]]> blocks (except in embedded mathml, but that's probably not what you're asking about) |
| 09:07 | <hsivonen> | at some point, someone needs to define case-sensitivity of role values |
| 09:08 | <hsivonen> | even if only to say that they aren't like other enumerated values |
| 10:07 | gsnedders | has proved once again on public-html he's an idiot |
| 10:19 | <hdworak> | Hixie: ok, thank you |
| 10:24 | <hdworak> | Hixie: so does that mean that from HTML 5 point of view, it is just ill-formed mark-up? |
| 10:30 | <Philip`> | hdworak: Yes - it's treated as a (bogus) comment, and the parser just reads up until the next ">" character after the "<!" |
| 10:30 | <hdworak> | Philip`: thank you |
| 10:31 | <Philip`> | so it should be parsed the same as <!--[CDATA[<sender-->John Smith</sender>]]> |
| 10:34 | <hdworak> | Philip`: ok, but I'm using your lib to parse any HTML/XHTML not just HTML 5, so I'm just going to translate all CDATA to text beforehand |
| 10:35 | <hdworak> | :) |
| 10:37 | <gsnedders> | hdworak: any HTML should be parsed per the HTML 5 rules |
| 10:37 | <Philip`> | hdworak: If it's actually really proper XHTML, you should probably use an XML parser instead; otherwise you should probably just use html5lib without any special translations, since it ought to match how browsers parse pages |
| 10:38 | <gsnedders> | (where proper XHTML is served as application/xhtml+xml, application/xml, or text/xml, NOT text/html) |
| 10:39 | <hdworak> | well, I'm then using a regex to match any <!-- <rdf:RDF ... comments, which are the deprecated way to embedded RDF/XML in the code |
| 10:39 | <hdworak> | if it's valid, I'm doing try: dom = minidom.parseString(code) |
| 10:39 | <hdworak> | first |
| 10:39 | <gsnedders> | hdworak: Where's the data coming from? |
| 10:39 | <hdworak> | if it fails, I'm doing html5lib |
| 10:39 | <hdworak> | direct input, file upload, remote URI |
| 10:40 | <gsnedders> | hdworak: For the first two, I'd make it user selectable. For the third, it should be done from Content-Type, optionally with content-type sniffing |
| 10:40 | <hdworak> | so, because I'm using regex to match this commented <rdf:RDF, I do need to take care of CDATA, because if this appears in CDATA, it should not be parsed at all |
| 10:41 | <gsnedders> | hdworak: de-facto CDATA doesn't exist in any HTML version, so that's irrelevant. What you need to know is whether the input is HTML or XML, nothing else. |
| 10:41 | <hdworak> | gsnedders: I understand, but IMHO the content-type really does not matter for this particular application |
| 10:41 | <hdworak> | I'm not doing a general purpose parser, but use it for extracting very specific information |
| 10:42 | gsnedders | notes that feeds got to the stage at one point where you absolutely had to completely ignore any content-type header because someone thought that |
| 10:42 | <hdworak> | I know about Content-Type, Content-Location, XHTML 1.1 required MIME, <?xml version="" encoding="", <meta content-type, xml:base, <base href="" |
| 10:43 | <hdworak> | I just do not think it does matter at all for this particular application |
| 10:43 | <gsnedders> | possible issue of ignoring it: what if I send Content-Type: text/html; charset=gb2312? |
| 10:43 | <hdworak> | I'll probably add content selection (HTML, XHTML, RSS) once I introduce RSS parsing |
| 10:43 | hsivonen | mumbles something about Trackback autodiscovery being so, so wrong |
| 10:44 | <hdworak> | gsnedders: does html5lib not return utf-8? |
| 10:44 | <gsnedders> | hdworak: You _really_ don't want to touch RSS parsing. Srsly. |
| 10:44 | <gsnedders> | hdworak: It wouldn't know what the input character encoding is. |
| 10:44 | <hsivonen> | I can imagine someone go "this is kinda like metadata. if you have metadata, you MUST use RDF. RDF won't validate. Let's comment it out" |
| 10:45 | <hdworak> | so it fails even for proper iso-8859-2 or something? |
| 10:45 | gsnedders | can't remember the details |
| 10:45 | <hdworak> | hsivonen: yes, but now they should use RDFa |
| 10:45 | <annevk> | Hixie, "requiring browsers to implementing rendering engines" |
| 10:45 | <hdworak> | does html5lib not try to guess the encoding? |
| 10:46 | <hsivonen> | hdworak: no, the point were the thinking goes wrong is at "if you have metadata, you MUST use RDF" |
| 10:46 | <hdworak> | like encutils or chardet? |
| 10:46 | <gsnedders> | hdworak: All it does is use meta@charset. |
| 10:46 | <gsnedders> | hdworak: and falls back to Windows-1252 |
| 10:46 | <hdworak> | huh... |
| 10:46 | <gsnedders> | hdworak: Oh, it does also use chardet |
| 10:46 | <hdworak> | so that's fine for the time being |
| 10:47 | <hdworak> | in the next versions I'll add content-type and encoding selection (aside of autodetection) |
| 10:47 | <hdworak> | like W3C Validator does |
| 10:47 | <hsivonen> | in retrospect, both Trackback and Pingback are more complex and bogus than they'd need to be |
| 10:47 | <hdworak> | but really, the first release does not have to be bulletproof |
| 10:47 | <hdworak> | pls understand |
| 10:47 | <Philip`> | To anyone who cares: http://wiki.whatwg.org/wiki/User_talk:Highhi and http://wiki.whatwg.org/wiki/User_talk:Hijing are spam |
| 10:48 | <hsivonen> | Philip`: I'll deal with those |
| 10:48 | <gsnedders> | hsivonen: Trackback needs more complexity insofar as to give a character encoding, really |
| 10:48 | <gsnedders> | hdworak: Plenty of things have become broken forever because the first release of a reader was broken (e.g., the web) |
| 10:49 | <hdworak> | :) |
| 10:49 | <hdworak> | I think this is going a bit too far |
| 10:49 | <hsivonen> | gsnedders: I think the right way to implement blong pinging would be to collect the link relation from Referer and for the pinged site to dereference the Referer URI and find the relevant <a href> element and take an extract around that |
| 10:49 | <gsnedders> | hsivonen: Yeah, that's Hixie's own opinion :) |
| 10:49 | <Philip`> | gsnedders: Plenty of things have failed even more badly because the first release of a reader was written too carefully and correctly and took too long to write and didn't have enough exciting features (e.g., the web's competitors) :-p |
| 10:50 | <hdworak> | thank you for your time for answering my questions, I do appreciate that, great channel |
| 10:50 | <gsnedders> | Philip`: Like me! I'm broken! |
| 10:50 | <Philip`> | (Those competitors have failed so badly that I can't even think of any specific examples, so I'm just assuming some must have existed) |
| 10:51 | <gsnedders> | Philip`: XHTML? |
| 10:51 | <Philip`> | gsnedders: XHTML isn't competing with the web |
| 10:52 | <gsnedders> | Philip`: True. But it's competing for the web. |
| 10:52 | <hsivonen> | Philip`: done |
| 11:00 | Philip` | finds a shop receipt that says "DVDs Coming Soon: ... NimÆs Island ...", and wonders where the character encoding issue got introduced |
| 11:02 | <Philip`> | hsivonen: Thanks |
| 11:03 | <virtuelv_> | hsivonen: you can mumble about trackback autodiscovery being wrong |
| 11:04 | <virtuelv_> | I'll just mumble about trackbacks being so wrong |
| 11:04 | <virtuelv_> | I had to turn support for them off globally on my sites, due to the insane amounts of spam that got through |
| 11:05 | <Philip`> | Is it the whole concept of bidirectional links that is wrong? |
| 11:06 | <virtuelv_> | no, rather the concept of untrusted bidirectional links |
| 11:08 | Philip` | isn't sure about the concept of trust either |
| 11:08 | <hdworak> | bye |
| 11:12 | <virtuelv_> | Philip`: there usually is none |
| 11:12 | <virtuelv_> | unless the trust is between two endpoints both controlled by the same entity |
| 11:13 | <virtuelv_> | or all approval of trust is manual, or (cryptographically) secure |
| 11:13 | <virtuelv_> | trackback fails in all such respects |
| 11:13 | <annevk> | man, Flickr Uploadr for the Mac sucks |
| 11:13 | <annevk> | keeps failing after a few images |
| 11:25 | hsivonen | uses his own upload script |
| 11:26 | <hsivonen> | annevk: do you want a copy? |
| 11:52 | <annevk> | hsivonen, yeah, why not |
| 11:52 | <gDashiva> | Philip`: Some cash registers here in Norway have a weird bug where lowercase norwegian letters become symbols, but uppercase ones work fine |
| 11:58 | <hsivonen> | annevk: emailed |
| 11:59 | <hsivonen> | annevk: you may treat the script as being under the MIT license |
| 12:04 | <annevk> | cheers |
| 13:45 | <gsnedders> | I am insane. |
| 13:46 | <gsnedders> | Four AHs, and a H. |
| 13:46 | <gDashiva> | Almost Honorable and Honorable? |
| 13:46 | <gsnedders> | Advanced Higher and Higher |
| 13:46 | <Philip`> | Ah |
| 13:46 | <gsnedders> | http://en.wikipedia.org/wiki/Advanced_Higher and http://en.wikipedia.org/wiki/Higher_(Scottish) |
| 13:48 | annevk | doesn't quite get the connection between insane and doing ok (at school, anyway) |
| 13:48 | <gsnedders> | annevk: No, that's what I'm doing this year. It is an insane amount of work. Nobody else in my year is doing it. |
| 13:48 | <gsnedders> | annevk: Very, very, very few people are doing four AHs anyway. |
| 13:48 | <gDashiva> | Maybe he's insane because he's not being consistent and doing 5 AH? |
| 13:50 | <gsnedders> | No, because I'm doing too much :) |
| 13:50 | <gsnedders> | But also because my name is gsnedders, therefore by definition I am insane. |
| 13:51 | <annevk> | makes perfect sense indeed |
| 13:52 | <gDashiva> | gsnedders: You have to rephrase your statement in lisp to qualify as patently insane |
| 13:53 | <gsnedders> | gDashiva: jfhaksdsfjkhdshjkdsf |
| 13:55 | Philip` | wonders who collects the royalties on the insanity patent |
| 13:55 | <gsnedders> | Philip`: Nobody — there's too much prior art |
| 13:57 | <gDashiva> | Are you implying artists are insane? |
| 13:58 | <gsnedders> | All art is insanity. |
| 14:00 | <jcranmer> | bah, I'm in my first class 95% of which will be review |
| 14:01 | gsnedders | goes back tomorrow |
| 14:01 | gsnedders | ought to type up a computing project proposal |
| 17:49 | <takkaria> | grr, oo.org spreadsheet is so far inferior to excel I don't know how anyone can switch from one to the other |
| 17:54 | <hsivonen> | takkaria: VBA in Excel is the big Windows lock-in these days |
| 17:56 | <takkaria> | I just mean the UI and graphing tools |
| 17:56 | <takkaria> | the kinds of things I used to use VBA for I can normally manage on the commandline |
| 19:17 | <gDashiva> | webapps is nick confusion waiting to happen... AnnB, ArtB, AdamB, and anne and arve_ to boot :) |
| 19:42 | <Philip`> | gDashiva: You should encourage them to each prepend a random letter to their name for disambiguation |
| 20:04 | <jgraham> | Adobe just sent me an email with no text but with a PDF attachment... containing nothing but the adobe logo and some text. |
| 22:13 | <Hixie> | wow |
| 22:13 | <Hixie> | just... wow |
| 22:13 | <Hixie> | fundamentalism has eached public-html in force. |
| 22:15 | <gavin_> | I find it kind of entertaining, actually |
| 22:15 | <gavin_> | I enjoy reading boris' arguments |
| 22:17 | <Hixie> | james, job, anne, and boris appear to be the voices of sanity in a sea of crazy |
| 22:17 | <annevk> | and hsivonen on #html-wg :) |
| 22:18 | <roc> | which thread are we talking about? |
| 22:18 | <Hixie> | it utterly baffles me that these people are being serious |
| 22:18 | <gavin_> | roc: "flickr and alt" |
| 22:22 | <Hixie> | oooh, squee! a win for the new event loop stuff. It allows a really easy fix for a problem that might have been a bitch to write up the fix for. |
| 22:24 | <roc> | heh |
| 22:27 | <roc> | once again I think the confusion is that people think HTML5 has the force of law |
| 22:28 | <roc> | maybe one of these people will go to court and get HTML5 non-conformance declared a violation of the ADA. Wouldn't that be exciting |
| 22:29 | <Hixie> | i hope i'm a witness in that court case |
| 22:29 | <Hixie> | so that i can help make that law pass |
| 22:29 | <Hixie> | i'll then carefully change html5 to require everytone to pay me $5 a day |
| 22:30 | <annevk> | did anyone see an e-mail about https://bugzilla.mozilla.org/show_bug.cgi?id=448674 ? |
| 22:31 | annevk | didn't |
| 22:32 | <roc> | it's completely obvious |
| 22:32 | <roc> | I'll send an email about it if you like |
| 22:33 | <annevk> | prolly good as it's not part of the spec now |
| 22:33 | <Philip`> | It's inconsistent with the spec's requirements for animated images (like animated GIF and APNG) |
| 22:33 | <Philip`> | (since the spec requires the poster frame or first frame to be rendered, I think) |
| 22:35 | <roc> | was there an email when Opera added SVG element support there? |
| 22:35 | <roc> | because I've got some questions about that :-) |
| 22:35 | <annevk> | yeah, I e-mailed |
| 22:35 | <roc> | ok cool |
| 22:35 | <annevk> | to public-html iirc |
| 22:35 | <annevk> | so might not have reached you, dunno |
| 22:35 | <roc> | oh |
| 22:35 | <roc> | dear |
| 22:36 | <roc> | hmm |
| 22:36 | <roc> | sigh |
| 22:39 | <roc> | thank goodness we're open source :-) |
| 23:02 | <Philip`> | "what's the poing?" - I have often wondered that |
| 23:05 | <annevk> | http://www.google.com/search?q=%22what's+the+poing?%22 |
| 23:08 | <_Xenos> | http://www.google.com/search?hl=en&hs=SkX&defl=en&q=define:poing |
| 23:08 | <_Xenos> | "a closed hand" seems to be the #1 definition |
| 23:14 | <gavin_> | heh |
| 23:14 | <gavin_> | that's french |